Introduction to Recursively Imputed Survival Trees

Apr 1, 2021 9:30 AM
MEB Biostatistics seminar, Karolinska Institutet

Abstract: Extrapolating relative survival (e.g., extrapolating log cumulative excess hazard using flexible parametric models) with more than 10 years of survival data has already been proved to give reasonable all-cause survival until lifetime. However, with data less than 10 years, it remains unsolved to obtain robust prediction. Recursively imputed survival trees adopted and modified the characteristics of random forest method, which aims to deal with censoring survival data and predict as much information as in non-censored data. Will this approach tackle the challenge of extrapolating survival with limited data?

Syllabus to the PDF!